Prompt Engineering Checklist 2025

One-page printable reference. Use with ChatGPT, Claude, and Gemini.

Before Writing

Goal: clear objective + success criteria
Audience: expertise level + tone
Deliverable: bullets, table, JSON, code, etc.
Constraints: scope, sources, length, tokens
Model: ChatGPT (structured/code), Claude (analysis), Gemini (multimodal/research)

Structure

Context: minimal background and data
Instructions: one explicit task; guardrails
Examples: 1–3 for style/format
Output: exact structure, tone, length

Optimize

BIV Loop: Baseline → Improve → Verify
Verification: assumptions, sources, edge cases
Rubric (0–5): Clarity, Evidence, Actionability
Save winners as templates; track cost

ChatGPT

Set role + headings/lists/tables
Split into numbered sub‑tasks
Few‑shot examples for consistency
Ask for JSON/code when integrating

Claude

Provide rich context + step‑by‑step
Request critique and uncertainty flags
Use structured sections for reasoning
Favor depth over breadth

Gemini

Call out multimodal/data inputs
Set research parameters + citations
Ask for verification + limitations
Prefer concise, scannable bullets

More from PromptBuilder

© 2025 PromptBuilder · promptbuilder.cc